Exons:
Relevant bits of information for using the MUSC Shared Resource,
the BioMolecular Computing Resource (BCR).

EXON_12:  Local Databases for FASTA and BLAST Searches

Building local databases.
At MUSC we receive the current bi-monthly release of Genbank 
and  the quarterly release of PIR via CDROM; other files are retrieved
 via FTP. The contents are then re-formatted  into the GCG searchable 
files. This process consists of a straightforward reformat (adding a pair 
of dots -see Exon_7). In addition the GCG software builds a series of 
indexes to allow the STRINGSEARCH of the databases. 

It may turn out that your research focuses on a class or a few classes 
of molecules/genes. Under these circumstances you might wish to 
consider building a local database which is specific for your molecule(s). 
For example you might build up a database of  cytochromes C, 
or immunoglobulins etc.  Such databases may be readily built from the 
GCG formatted databases by using one of two utilities and several 
search programs.

The first step is to create a file of sequence names. This may of course 
be done in a tedious manual fashion. Alternatively, you might wish to 
make use of the output of the GCG stringsearch program or the 
wordsearch program. In the former, you are seeking to find particular 
character strings in the header fields of the sequence files of the 
archive. For example you might do a search for "homeobox" genes. Any 
sequence with the string "homeobox" would cause a record of 
pointers to that sequence to be written to an output file. 
Alternatively,you might wish to find a series of sequences which
show sequence similarity to a query sequence of yours. WORDSEARCH 
can find similar sequences and writes the pointers to such 
sequences to an output file.  The output files may be easily edited
to add or remove sequences. For example a PIR search may reveal 
several proteins of a class but a nucleic acid search may reveal 
additional genes whose transcription products have not yet 
been added to the PIR archive. By manually translating the gene 
and manually editing the PIR stringsearch output file you may 
create a more inclusive listing.

The next step is to use the output file or the edited output file as
input for the utility program called "dataset". For instance, assume 
you have a file called 'homeobox_new.strings' which you wish to turn 
into a searchable database.Issue the command "dataset" and point
 the program to your file of sequence names, homeobox_new.strings. 
Like this:

%dataset @homeobox_new.strings

It's that easy.  DATASET generally takes only a few minutes to run  
so updates are quite painless. The program does ALL the rest. You 
are prompted for a dataset name and that is as hard as you have to 
work. The program creates about 8 files, all with the same prefix (ie 
the name of your database).  To make use of your new database you 
must specify it rather than a subset of GCG, PIR etc.
Here's how to do that. Assume you called your database "hombox"
When prompted for a set of sequences to search, reply with 

hombox:*

And  the program (stringsearch, or more likely FASTA or 
WORDSEARCH) uses your new local database!

The GCG v8.0 allows BLAST searches of local databases.
These databases are created with the program "TOBLAST". 
Notifying the system of the existence of your local database requires
 that you complete several tasks.

1. Create the database with the "toblast" program. TOBLAST can 
create a 'blastable" database (in a binary form) from any list of 
sequences. Output from WORDSEARCH and STRINGSEARCH are 
particularly useful input  files for TOBLAST.

2. Fetch (a GCG command) the file called "blast.sdbs". Type 
"fetch blast.sdbs". The system then writes the ascii file 
blast.sdbs to your current directory (it is advisable to write 
the file to your personal root directory).

3. Edit the blast.sdbs file to point to your newly minted local 
blastable database(s) by adding a line which reads includes a path,
a name, a type (eg protein or nucleic acid), and a comment. 
Example lines:

/home/hazards/project1/ides n A nucleotide database
/home/hazards/project2/ICE p A protein database

You must add a name for each local database you create!

4. At the time you evoke the GCG BLAST interface you must add the 
option
"-dat3=/path/blast.sdbs"
This tells the system where to find the file that points it to
ALL of your local databases.

After indicating your  query sequence you will see a screen of 
databases preceded by numbers. At the bottom of the screen it 
will say "local databases" and below that will be numbers of  
each of your databases. Selecting the number of one of your 
local databases will direct the GCG interface to use your query 
sequence and search the indicated local database.